"Reading with AI," episode 003. The Intelligent Investor (4th edition, 1973, with Jason Zweig's commentary), Chapter 3: "A Century of Stock-Market History: The Level of Stock Prices in Early 1972." The first two chapters dealt with "are you investing or speculating" and "will inflation bail stocks out." In this chapter, Graham does something almost nobody is willing to do today: he lays out 100 years of history since 1871, the highs and lows of 19 bull-and-bear cycles, and the decade-by-decade relationships among stock prices, earnings, and dividends—then gives "is the market expensive right now" a computable answer. Even more worth reading is what he does immediately afterward: he undercuts himself, publishing his own report card from four market judgments made in 1948, 1953, 1959, and 1964—including the one that "subsequent experience proved was not a particularly good recommendation."

A century of stock-market history can't tell you whether prices rise or fall tomorrow, but it can hand you a ruler for measuring "expensive versus cheap"

The Core Question This Chapter Answers

How should an investor use stock-market history? Concretely: once the long-term relationships among stock prices, earnings, and dividends are laid out, can you arrive at a valuable judgment about "the attractiveness and dangers of the current market level"?

Graham states his position plainly at the outset: the portfolio of common stocks owned by an investor is but a small cross-section of the vast organism that is the stock market. As a matter of prudence, he ought to know something about market history—particularly the major fluctuations in prices and the various relationships between overall price levels, earnings, and dividends—and on that basis he should be able to form some valuable judgment about the attractiveness or danger of the price levels of different periods.

Note that the words "predict" or "forecast" appear nowhere in that sentence. History's use is to locate the present price, not to schedule the future. That is the methodological premise of the whole chapter and the skeleton of everything that follows.

---

Graham's Text

Key Point 1: Why Start in 1871?—Stating the Boundaries of the Data Up Front

Graham is candid about it: fairly complete statistical data on stock prices, earnings, and dividends begin in 1871, 100 years earlier, and "the data for the first 50 years are not as complete and reliable as those for the last 50, but they can still be used."

He uses two indices:

Index	Source and composition
Cowles Commission index	Begins in 1870; the forerunner of the S&P 500, forming one continuous whole with it
Dow Jones Industrial Average (DJIA)	Born in 1897; 30 component stocks—29 large industrial companies plus one, American Telephone & Telegraph

This accounting of "where the data comes from and how reliable it is" matters a great deal—Zweig will pick it up later and turn it into a blade to cut down the 1990s bestsellers claiming "stocks have returned 7% a year in real terms for 200 years." Keep this foreshadowing in mind.

Key Point 2: 70 Years, Cut into Three Segments, Each with a Completely Different Annualized Return

Figure 3-1 shows the S&P industrial index (425 industrial stocks) from 1900 to 1970. Graham asks the reader to notice one thing: these 70 years fall into three phases, each spanning roughly a third of the period, and their average annual gains differ enormously.

Phase	Character	Average annual gain
1900–1924	A series of 3-to-5-year market cycles with strikingly similar behavior	about 3%
1924–1949	The bull-market "New Era" → the 1929 peak → the Great Crash → irregular fluctuations to 1949	only 1.5%
1949–1968	The greatest bull market in history	see below

The sentence describing the end of the second phase deserves to be copied out: "Thus, at the end of this period, the public had no interest whatsoever in stocks. By the law of action and reaction, the time was ripe for the greatest bull market in history."

That idea—the law of action and reaction—is the hidden thread of this chapter, and it will return in another form at the end to slam into the optimists.

The third phase in numbers: from 163 in mid-1949 to 995 in early 1966, the Dow rose fivefold over 17 years, an 11% compound annual gain, plus roughly 3.5% per year in dividends (the S&P 500 did even better, going from 14 to 96). The move probably topped out in December 1968 (118 on the S&P 425 industrial index / 108 on the S&P 500). There were substantial setbacks in 1956–1957 and 1961–1962, but by the long-standing convention these count as reactions within the same bull-market cycle, not separate cycles.

Key Point 3: The 1963 "14% Return Record"—and "a Dangerous Conclusion That Made No Sense"

This is the first key scene of the chapter, and the one today's readers should most readily recognize themselves in.

In 1963 there appeared a return record in excess of 14%, which was then written up in a widely noted academic study (the classic paper by Fisher and Lorie, "Rates of Return on Investments in Common Stocks: The Year-by-Year Record, 1926–65," in the Journal of Business, July 1968).

Naturally, Wall Street was pleased with this fine achievement and expected these high levels of return to continue into the future—a dangerous conclusion that made no sense at all. Few considered that the advance had already carried too far.

The consequence: from the 1968 high to the 1970 low, the S&P fell 36% and the Dow 37%. For comparison, the previous largest decline occurred in 1939–1942, a 44% drop reflecting the risks and uncertainties after Pearl Harbor.

Put those two numbers side by side: one was a world-war-scale panic (–44%), the other merely "everyone believed high returns would continue" (–37%). Overvaluation alone can manufacture a decline approaching wartime magnitude.

And here is Wall Street's gift for theater: both indices rebounded rapidly from their May 1970 lows, and the S&P industrial index even reached an all-time high in early 1972. That is the scene as Graham writes—a crash just happened, then a new high, and people had already forgotten.

He adds a comparison that is easy to overlook: from 1949 to 1970 the S&P grew about 9% a year, far above any comparable period before 1950; but the final 10 years were much weaker—the S&P 500 averaged 5.25% a year, the Dow only 3%.

Key Point 4: Table 3-2 Is the Chapter's Real Tool—Decade Averages of the Three-Way Relationship

Prices alone are not enough. Graham writes: "To derive a full picture of the economics of stocks over the past 100 years, one must know about earnings and dividends as well as prices." Table 3-2 uses ten-year averages to smooth out year-by-year fluctuations, giving prices, earnings, P/E ratios, dividends, dividend yields, payout ratios, and growth rates.

A few selected decades (S&P composite basis):

Period	Average price	Average earnings	P/E ratio	Dividend yield	Payout ratio
1871–1880	3.58	0.32	13x	6.0%	67%
1891–1900	4.65	0.30	15.5x	4.0%	66%
1911–1920	8.62	0.86	10.0x	5.8%	58%
1931–1940	11.55	0.68	17.0x	5.1%	85%
1941–1950	13.90	1.46	9.5x	6.3%	60%
1951–1960	39.20	3.00	13.1x	4.2%	54%
1961–1970	82.50	4.83	17.1x	3.2%	55%

His reading has three layers:

1. Overall, the picture is one of continuous advance. In the nine decades after the first, only two saw declines in the average level of both earnings and prices (1891–1900 and 1931–1940); and after 1900, the average dividend of each decade never declined.

2. But growth rates varied greatly. The postwar decades were generally stronger than the prewar ones, yet the growth rate of the 1960s was lower than that of the 1950s.

3. Hence the conclusion must be restrained to the point of deflating:

Today's investor cannot see from these records how much he may expect to gain over the next 10 years from rising dividends and stock prices, but they do provide sufficient guidance for a consistent policy of stock investment.

That sentence is the chapter's credo: the data give you no forecast, only a reason for a "consistent policy."

He also singles out one problem the table fails to reveal: in 1970 the earnings of American corporations as a whole deteriorated markedly, and the return on invested capital fell to its lowest level since World War II; a considerable number of companies operated at a loss that year, many fell into "financial difficulties," and corporations entering bankruptcy proceedings reached the highest number in nearly 30 years. The conclusion interlocks with Chapter 2: the era of great prosperity came to an end in 1969–1970.

Key Point 5: "The Most Striking Shift in Public Attitude" Ever Recorded

Then comes a set of contrasts—minimal in form, brutal in effect. These are the numbers in this chapter most worth memorizing:

	June 1949	March 1961
S&P composite P/E ratio	6.3x (on trailing 12-month earnings)	22.9x
S&P dividend yield	over 7%	only 3.0%
High-grade bond interest rate	2.60%	4.50%

This was unquestionably the most striking shift in public attitude toward the stock market that has ever taken place.

Notice the presence of the third row—as stocks became more expensive, the alternative (bonds) actually got better. This continues Chapter 2's point that "the attractiveness of any asset is relative."

Incidentally, in a footnote Graham supplies a crude but extremely useful P/E yardstick (his own words, "a simple tool for measuring the state of the market"): a P/E under 10 is low, 10 to 20 is moderate, and above 20 is considered excessive.

And he honestly records the fact that "the feared troubles never materialized": to people of long experience and cautious temperament, a market jumping from one extreme to the other is an ominous sign of serious trouble; they would remember 1929 with dread—but the feared troubles did not materialize. In fact, the Dow's closing level in 1970 was about the same as six and a half years earlier; the much-touted "soaring Sixties" had amounted to a series of small advances followed by declines, and neither business nor stock prices had produced anything comparable to the great bear market of 1929–1932.

The very existence of this passage is a form of honesty: he did not dress up "prudence" as "being right every time."

Key Point 6: Self-Judgment—The Published Report Card of Four Calls

This is the rarest kind of writing in the entire book: the author drags out his market judgments from successive editions and reviews them one by one, including the wrong one.

Year	Dow	His judgment at the time	Afterward
1948	180	"Measured by intrinsic value, stock prices at that time were not high"	Sound
1953	275	"Acceptable from the standpoint of value content," but he worried the advance already exceeded most historical bull markets and the absolute level was a record high; he advised investors to act with caution and adopt a balanced policy	"Subsequent experience proved this was not a particularly good recommendation"—the market rose another 100% over the following five years
1959	584	"The present level of stock prices is quite dangerous... even if it were not so, the market's own momentum would carry it to unreasonable heights"	Slightly better, but still far from what followed
1964	892	"Frankly, if the 1964 index is not too high, then no price level is too high"	The Dow rose another 11.9% to 995, then fell to 632 in 1970, closing the year at 839. "This time, caution proved justified."

The wording of the 1953 self-criticism is especially fine: "A good prophet should have foreseen that the market would rise 100% over the next five years. Perhaps we may offer this self-defense: among those engaged in forecasting the stock market, almost no one held views about the future that were more logical than ours."—Being logically right ≠ being right on timing, and he himself is the specimen.

The 1962 episode is the most vivid footnote of all: the Dow fell from 735 in December 1961 to 536 in May 1962—a drop of 27.9% in barely six months; the widely favored "growth stocks" fell hardest, and the bellwether IBM slid from $607 in December 1961 to $300 in June 1962; meanwhile a batch of high-priced "hot issues," driven to insane levels by the speculative wave, collapsed—many lost 90% or more within a few short months.

Then comes the complete cycle of sentiment, worth reading line by line:

• At the June 1962 low: predictions about the market's future were predominantly bearish;
• After the market recovered part of the loss by year-end: they turned mixed—with a tendency toward skepticism;
• Beginning in 1964: optimism at brokerage firms surged again, and predictions about the market's future were almost unanimously bullish, remaining so as the market rose.

This is the documentary version of "Mr. Market" (a preview of Chapter 8): the direction of forecasts always follows the price that just happened.

Key Point 7: The Concrete 1964 "Which Way Now" Checklist

In assessing November 1964 (Dow 892), he gave three conclusions, the first of which reads as strikingly familiar today:

1. "The old standards (of value) no longer seem applicable, and the new ones have not yet stood the test of time."
2. Investors "must plan their policy in the light of certain major uncertainties," and must consider both extremes: a further 50% rise (Dow 1,350) or an equivalent collapse (Dow around 450).
3. "Frankly, if the 1964 index is not too high, then no price level is too high."

And the actions he proposed were four directives you could copy verbatim:

1. Do not borrow money to buy or hold securities.
2. Do not increase the proportion of funds used to purchase stocks.
3. Reduce the stock position to below 50% of total investment; take full advantage of capital-gains tax benefits where possible, and put the money into the highest-grade bonds or savings deposits.
4. Investors following a dollar-cost averaging plan may continue to buy regularly, or they may change the approach when market prices feel safer.

After item 4 comes a supplement of remarkable psychological insight: "We would strongly advise against inaugurating a new dollar-cost averaging plan at the price levels prevailing at the end of 1964, because if serious adversity occurred shortly after such a plan was begun, its adherents would not continue with the practice."

Note the type of reasoning here: the objection to starting a new dollar-cost averaging plan at high levels is not that the arithmetic fails, but that novices abandon the plan when it opens with a huge loss. The same strategy has completely different success rates for veterans and beginners—this is a model of building behavioral risk into strategy design.

Nor did he forget to leave readers an exit: "Investors should not conclude that the price level of 1964 is too high merely on the basis of what this book says; they should compare our argument with the contrary arguments of other seasoned experts on Wall Street. In the end, everyone must make his own decision and take responsibility for it."

Key Point 8: The Actual Valuation in Early 1972—How He Did the Math

Now the main event. At various points in 1971 the Dow stood near 892, the same level as November 1964, discussed in the prior edition. This time he switches to the S&P composite index (more comprehensive and representative than the 30-stock Dow) and rounds the current level to 100.

His quotation from Aristotle's Ethics is a statement of his attitude toward "precision in valuation":

"It is the mark of a trained mind to rest satisfied with the degree of precision that the nature of the subject permits, and not to seek exactness where only an approximation is possible." Financial analysis lies somewhere between the mathematician and the orator.

The core data of Table 3-3 (including 1971; bond yields and wholesale prices are as of August of that year):

	1948	1953	1958	1963	1968	1971
Closing price	15.20	24.81	55.21	75.02	103.9	100 (≈Dow 900)
Current earnings	2.24	2.51	2.89	4.02	5.76	5.23
Trailing 3-year average earnings	1.65	2.44	2.22	3.63	5.37	5.53
Current dividend	0.93	1.48	1.75	2.28	2.99	3.10
High-grade bond yield	2.77%	3.08%	4.12%	4.36%	6.51%	7.57%
Price / 3-year earnings	9.2x	10.2x	17.6x	20.7x	19.5x	18.1x
3-year earnings yield (earnings/price)	10.9%	9.8%	5.8%	4.8%	5.15%	5.53%
Dividend yield	5.6%	5.5%	3.3%	3.04%	2.87%	3.11%
Stock earnings yield / bond yield	3.96x	3.20x	1.41x	1.10x	0.80x	0.72x
Dividend yield / bond yield	2.1x	1.8x	0.80x	0.70x	0.44x	0.41x
Earnings / book value	11.2%	11.8%	12.8%	10.5%	11.5%	11.5%

The chain of reasoning is exceptionally clean—three steps:

Step one (looks like good news): In October 1971 the trailing three-year P/E ratio was lower than at year-end 1963 and 1968, and roughly in line with 1958—but far higher than in the years when the great bull market began.

Step two (bring interest rates in): These indicators by themselves do not show that stock prices in January 1972 were too high, but if the yield on high-grade bonds is brought in, the picture is not so reassuring. Compared with bond returns, stock returns in this period had deteriorated. The most glaring item is the complete reversal in the last two rows:

In 1948 the dividend yield was twice the bond yield; now the bond yield is twice the dividend yield, or more.

Step three (the conclusion):

On the basis of three-year average earnings, the reversal of the bond-to-stock yield ratio is enough to offset the decline in stock P/E ratios by the end of 1971. Hence our view of the price levels of early 1972 is exactly the same as it was seven years ago: from a conservative investment standpoint, valuations at this point lack attractiveness.

This is the complete demonstration of "relative valuation": the P/E came down, but the conclusion did not—because the risk-free alternative you could get had improved by even more. Expensive and cheap has never been an absolute number; it is a ratio.

Key Point 9: "Will This Carelessness Escape Punishment?"

In closing, he performs a stratification of risk—a fine lesson in how to say "I don't know which way, but I know what to guard against."

On the one hand, he concedes there is no reliable danger signal: the 1971 market still seemed to be in an unstable recovery phase after the sharp 1969–1970 decline; similar recoveries in the past had led to enduring bull markets, as after 1949 (and Wall Street in 1971 was full of that expectation); and given how many buyers had been badly burned purchasing low-grade new issues in 1968–1970, it was clearly too early for a new wave of new-issue speculation in 1971. So "the market has not yet given any dependable signal of impending danger," and strictly speaking, before the next serious setback or crash, the Dow would seem to have to move substantially above the 900 level once again.

On the other hand, he refuses to look away:

In our view, the market's behavior in early 1971—ignoring the painful lesson of less than a year before—is itself a disturbing sign. Will this carelessness escape punishment? We believe the investor must think ahead about the difficult period to come—perhaps a rapid repeat of the 1969–1970 collapse; perhaps first a powerfully bullish run-up, followed by a disastrous crash.

Feel the force of that construction: he does not predict which path; he writes both down and demands that your portfolio survive either. It is the same grammar as Chapter 2's "precisely because the future is uncertain."

The answer to "which way now" is a single sentence: When the Dow returns in early 1972 to the same level as at the end of 1964—around 900—our view is unchanged.

---

Jason Zweig's Commentary (Fourth Edition)

He opens by quoting Yogi Berra: "You've got to be very careful if you don't know where you're going, because you might not get there."

Commentary 1: He Saw 1973–1974 Coming

Zweig first credits Graham: looking two years ahead, he successfully foresaw the "disastrous" bear market of 1973–1974, during which U.S. stocks fell 37% (footnote: excluding dividends, the two-year drop was 47.8%). And his gaze reached through the next 20 years, completely exposing the tricks of today's market gurus and bestselling authors—even though those tricks had not yet appeared in Graham's day.

The core principle, violated again and again: the intelligent investor must never forecast the future by extrapolating the past. That is precisely the mistake the learned men of the 1990s made over and over.

Commentary 2: The Book Titles Alone Are Alarming—From "Dow 36,000" to "100,000"

Following the 1994 publication of Stocks for the Long Run by Jeremy Siegel, a professor of finance at Wharton, came a string of similar books, the most conspicuous of which (all published in 1999) were:

• James Glassman and Kevin Hassett, Dow 36,000;
• David Elias, Dow 40,000;
• Charles Kadlec, Dow 100,000.

The prophets' argument: since 1802, stocks have returned an average of 7% per year after inflation; therefore investors may expect the same in the future.

Some bull-market advocates went even further: since over the past 30 years stocks had "always" beaten bonds, they must be less risky than bonds—and even less risky than cash in the bank; if holding long enough eliminates all risk, why should anyone care about the purchase price in the first place?

The three on-record remarks Zweig collected are a perfect slice of the era:

Date	Person	Quote	Outcome
1999-12-07	Kevin Landis, manager of Firsthand Funds (CNN Moneyline, asked whether wireless-telecom P/E ratios were overvalued)	"That's not excessive. If you look at their rapid growth, the absolute value of that growth is enormous."	His favorite holding, Nokia, fell "only" 67% from 2000 to 2002; his worst, Winstar Communications, fell 99.9%
2000-01-18	Robert Froehlich, chief investment strategist at Kemper Funds (The Wall Street Journal)	"This is a new world order.... Because share prices are so high, people are selling the stocks of perfectly good companies. It's the dumbest mistake investors could make."	His top picks Cisco and Motorola plunged more than 70% by 2002; investors lost $400 billion on Cisco alone—more than the combined GNPs of Hong Kong, Israel, Kuwait, and Singapore
2000-04-10	Jeffrey Applegate, strategist at Lehman Brothers (BusinessWeek)	"Is the stock market riskier just because prices are higher than they were two years ago? The answer is no."	When he asked, the Dow was at 11,187 and the Nasdaq at 4,446; by the end of 2002 the Dow hovered around 8,300 and the Nasdaq had retreated to 1,300, wiping out the entire six-year gain

But the answer is yes. It was in the past, and it always will be.

When Graham asked, "Will this carelessness escape punishment?", he knew the answer is forever no. Like an enraged Greek god, the stock market wrecked everyone who believed the high returns of the late 1990s were justified in any way.

Commentary 3: "Survival of the Fattest"—That 7% Figure Is Itself Bogus

This is the most intellectually substantive passage of the chapter's commentary; it kicks the floor out from under "stocks always beat bonds in the long run."

The claim that stocks "always" beat bonds over the long run has a fatal flaw: reliable data before 1871 do not exist.

How big is the flaw, specifically?

• The index used to represent the earliest returns of the U.S. stock market contained just 7 stocks (Zweig adds, emphatically: yes, only 7);
• Yet by 1800, the United States had roughly 300 companies (many the Jefferson-era equivalents of the internet—wooden-cartwheel makers, canal developers);
• Many of those companies later went broke, and their investors went down with them;
• But the stock indices never included those early failures—the so-called "survivorship bias."

The result: these indices drastically overstate what investors actually earned—since no one had perfect foresight about exactly which 7 stocks to buy. For every miraculous survivor (like the Bank of New York or J.P. Morgan), no fewer than a thousand companies met financial doom—including the Dismal Swamp Canal Company, the Pennsylvania Vineyard Company, the Snickers Gap Turnpike Company, and on and on—none of which appear in the "historic" stock indices.

The numerical correction is devastating:

	Annual return, 1802–1870 (after inflation)
Siegel's data: stocks	7%
Siegel's data: bonds	4.8%
Siegel's data: cash	5.1%
Correction by Timson et al. of London Business School	stock returns overstated by at least two percentage points

Thus, in the real world, stocks did not outperform bonds and cash—and may have done worse. Only the ignorant can claim that history has "proven" stocks will always beat bonds and cash.

(The two footnotes matter too: not until the 1840s did these indices expand to 7 bank stocks plus 27 railroad stocks—"an absurd and unrepresentative sample for the young American market"; and stocks from 1871 through the 1920s remain affected by survivorship bias—hundreds of automobile, aircraft, and radio companies in that period died halfway and vanished into history, so returns for that stretch may also be overstated by two percentage points.)

This passage echoes Graham's opening line that "data before 1871 are incomplete"—the same data boundary that Graham used to restrain his conclusions, and that bestselling authors used to inflate their promises.

Commentary 4: Michael Jordan and "The Value of Any Investment Depends on the Price You Pay"

Graham's answers always came from logical analysis and common sense:

The value of any investment is, and always will be, a function of the price you pay for it.

By the late 1990s inflation was dormant, corporate profits were booming, and most of the world was at peace—but that did not mean, and never will mean, that stocks are worth buying at any price. Since the profits a corporation can earn are finite, the price an investor pays for them should have a limit.

The analogy is a stroke of genius:

Michael Jordan may be the greatest basketball player ever, a giant magnet pulling fans into Chicago Stadium. The Bulls paid him as much as $34 million a season for this—but that does not mean the Bulls should have paid him $340 million, $3.4 billion, or $34 billion per season.

"Great" and "worth the price" are two different questions—this one line can settle 90% of today's arguments about "a great company is worth buying at any price."

Zweig also offers three questions in the spirit of "good, long-acting medicine":

1. Why should the future return of stocks always be the same as their past return?
2. If every investor is certain that stocks are guaranteed to make money in the long run, won't stock prices be driven to severely overvalued levels by that very belief?
3. And if that has already happened, how could future returns possibly be high?

Commentary 5: The Limits of Optimism—Expectations Follow Prices, Not the Other Way Around

From 1995 to 1999 the U.S. market rose more than 20% every year—unprecedented in American history. And so:

Point in time	Expectations of investors/institutions
Mid-1998	Gallup poll for PaineWebber: investors expected average stock returns of about 13% over the next year
Early 2000	Expected returns surged to more than 18%
2001	SBC Communications raised its pension return assumption from 8.5% to 9.5%
2002	S&P 500 companies lifted the average expected return of their pension plans to a record 9.2%
2001–2002	Gallup: investors' expected one-year return plunged to 7%—even though they were now buying stocks at half the prices of 2000

The bill came due: by Wall Street's own estimates, companies' over-optimistic pension assumptions would cost them at least $32 billion.

While everyone knows they should buy low and sell high, the actual result is usually the opposite. "By the rule of action and reaction," the more bullish investors feel about the long-term market, the more likely they are to be wrong in the short run.

The closing numbers: on March 24, 2000, total U.S. stock-market value peaked at $14.75 trillion; just 30 months later, on October 9, 2002, it had fallen to $7.34 trillion—a 50.2% decline, with $7.41 trillion of market value evaporating. Meanwhile, many market pundits had turned deeply pessimistic, predicting flat or negative returns for years or even decades to come.

Hence the Grahamian question: given how badly "the experts" botched their last forecast, why should the intelligent investor believe them now?

This is the complete empirical demonstration of "expectations are a lagging indicator of prices": at full price, expectations of 18%; at half price, expectations of 7%—exactly backwards.

Commentary 6: Estimating Future Returns Yourself—The Three-Factor Addition

Zweig offers a framework you can still use directly today. The market's course depends on three factors:

1. Real growth (rising corporate profits and dividends);
2. Inflationary growth (a general rise in prices);
3. The growth or shrinkage of speculative activity (the investing public's rising or falling interest in stocks).

Plug in the numbers from early 2003 (long-term earnings per share growing 1.5%–2% a year before inflation; inflation around 2.4%; dividend yield 1.9%):

Real growth        1.5% ~ 2%
+ Inflation             2.4%
+ Dividend yield        1.9%
= Reasonable expectation   5.8% ~ 6.3%

Over the long run, this means a reasonable expectation for annual stock returns should be around 6% (or about 4% after inflation). If the public turns greedy again and drives stocks into an uptrend, that speculative enthusiasm will push returns briefly higher; conversely, if investors are gripped by fear (as in the 1930s and 1970s), stock returns will suffer a temporary decline.

Note where the third factor sits: speculative sentiment can produce only "brief" and "temporary" deviations—the foundation of long-run returns is the first two factors plus dividends. This is the same idea as the Gordon equation in Chapter 1, written in a different notation.

Commentary 7: The Shiller P/E—The Modern Version of Graham's Method

Robert Shiller, professor of finance at Yale, says his valuation method derives from Graham: compare the current level of the S&P 500 with companies' average earnings over the past 10 years (adjusted for inflation).

The verdict of historical testing is plain: when this P/E ratio has risen above 20, subsequent market returns have typically been low; when it has fallen below 10, subsequent returns have been excellent.

By Shiller's arithmetic: in early 2003, stock prices stood at 22.8 times companies' average inflation-adjusted earnings of the previous 10 years—still in the danger zone, though well down from 44.2 in December 1999.

So what happened over the following 10 years when history stood at similarly high levels? The most representative rows from the table (P/E / average annual total return over the following decade):

Year	Shiller P/E	Average annual total return, next 10 years
1929	22.0	–0.1%
1936	21.1	4.4%
1955	18.9	11.1%
1964	22.8	1.2%
1965	23.7	3.3%
1968	22.3	3.2%
1972	18.6	6.7%
1992	20.4	9.3%
18-sample average	20.8	6.0%

Thus, from valuation levels in early 2003 similar to those in the past, the stock market's performance over the following decade was sometimes excellent, sometimes dreadful, and sometimes mediocre.

Zweig's inference is a nice one: true to his consistently conservative temperament, Graham would have averaged the best and worst past returns and forecast an annual stock return of 6%—or 4% after inflation—over the coming decade. (Interestingly, that forecast matches exactly the number produced by the three-factor addition above.)

Compared with the 1990s, 6% is small change, but somewhat better than what bonds were likely to produce—and for most investors, that alone justified keeping stocks in their portfolios.

(Note the 1964 row: Graham's judgment that year was "if the 1964 level is not too high, then no level is too high." Shiller's data gives him a perfect score: 22.8x, and a subsequent decade averaging 1.2% a year.)

Commentary 8: The Second Lesson—The Only Certainty Is That You May Be Wrong

About forecasts of future stock returns, the only certainty is that you may well reach the wrong conclusion. The only indisputable truth history teaches us is that the future will always surprise us—always.

One corollary of this law of financial history: those who are most surprised by the stock market are precisely those most confident in their forecasts. Humility, like Graham's, will spare you the agony of being sure of yourself and losing everything anyway.

The closing passage is gentle, and important: So keep your expectations as low as you can—without becoming discouraged. To the intelligent investor, hope springs eternal, because it should. With the stock market, the worse the future looks, the better it usually turns out to be.

Finally, the Chesterton exchange—a cynic said, "God will bless the man who expects nothing, for he shall never be disappointed." Chesterton replied:

"God will bless the man who expects nothing, for he shall enjoy everything."

---

Untangling the Hard Parts

Hard part one: What exactly should you "learn" from a century of market history?

Not "what falls far will rise again." Graham's usage has three rules, each more concrete than the last:

1. Learn to segment: 1900–1924 annualized 3%; 1924–1949 annualized 1.5%; 1949–1968 annualized 11%. The same market, a different personality every couple of decades—so any "long-run return" extrapolated from a single stretch of history carries a built-in sampling bias.
2. Learn the three-way relationship: prices alone mean nothing; you must look simultaneously at earnings, dividends, P/E ratios, dividend yields, and bond yields. The two rows in Table 3-3—"stock earnings yield/bond yield" and "dividend yield/bond yield"—are where he actually renders judgment.
3. Learn discipline, not timing: he himself was cautious in 1953, and the market then rose another 100%. History hands you probabilities and a sense of discipline, not a timetable; the market can stay "irrational" longer than you can stay solvent.

Hard part two: Why did the P/E fall by 1971, yet the conclusion remained "unattractive," same as 1964?

This is the step in this chapter most worth thinking through. In 1971 the trailing three-year P/E of 18.1x was actually lower than 1963's 20.7x; but over the same period the high-grade bond yield soared from 4.36% to 7.57%.

So both ratios deteriorated across the board: stock earnings yield/bond yield fell from 1.10x to 0.72x; dividend yield/bond yield fell from 0.70x to 0.41x. In Graham's words, "the reversal of the bond-to-stock yield ratio is enough to offset the decline in stock P/E ratios."

Conclusion: "expensive or not" is not about the absolute P/E; it's about how good the alternative you can get is. The moment interest rates move, the attractiveness ranking of all assets is reshuffled—this is Chapter 2's "the attractiveness of any asset is relative" making its second appearance in this chapter, and it is the step most often skipped when people judge markets today.

Hard part three: Where exactly does "stocks always beat bonds in the long run" go wrong?

It fails on two levels:

• The data level (survivorship bias): before 1871 the sample was only 7 stocks, while by 1800 the United States already had about 300 companies—the failures never entered the index. After correction, stock returns for 1802–1870 were overstated by at least two percentage points—in reality stocks did not beat bonds and cash, and may have done worse. Never mind the hundreds of failed automobile, aircraft, and radio companies of 1871 through the 1920s, likely another two-point overstatement.
• The logic level (self-refutation): if everyone becomes certain that holding stocks long enough must pay, prices will be driven to a level that makes future returns low. So the moment the belief "stocks always win in the long run" spreads widely, it begins destroying its own premise.

The correct formulation: the long-run return of stocks is worth counting on, but its size depends on the price you pay; and as for "history has proven stocks always win"—history itself never said that.

Hard part four: How to use the three-factor addition today?

The framework is dividend yield + real earnings growth + inflation = expected long-run annual return, with speculative sentiment affecting only the short run. The key to using it is filling in the three numbers honestly: the dividend yield is easy to look up; real earnings growth should be a long-term average (Zweig uses 1.5%–2%), not the last three explosive years; inflation should be a sustainable level.

If your computed expectation is far above 6%–8%, the excess can only come from the third factor—which means you are betting that "someone later will pay a higher price." This translates an idea like "I expect 20% a year" into a statement that can actually be tested.

Layer on the Shiller P/E as a cross-check (>20x historically leads to lower subsequent returns; <10x to excellent ones; the 18 high-level samples averaged 6.0% over the following decade), and you have Graham's two rulers: one measures the absolute level, one measures the level relative to bonds.

The One-Line Takeaway

History is not for extrapolating; it is for locating yourself: put prices, earnings, and dividends together with bond yields, and only then do you know whether the market is expensive—by 1971 the P/E had clearly fallen, yet the conclusion was still "unattractive," because bond yields had overtaken dividend yields by more than two to one. The value of any investment always depends on the price you pay; however great Jordan was, he was not worth $34 billion a season. And the notion that "stocks always win in the long run" is an illusion inflated by two percentage points of survivorship bias. The more certain you are of the long-term bull case, the more likely you are to be wrong in the short run.

Questions to Leave You With

1. Give your own market a checkup using Graham's Table 3-3 method: index dividend yield vs. the 10-year Treasury yield—which is higher? What's the ratio? Then compare with 1948 (dividend yield twice the bond yield) and 1971 (bond yield more than twice the dividend yield)—which year are you closer to now?
2. Run the three-factor addition for your own reasonable expectation over the next decade: dividend yield + long-term real earnings growth + inflation. How far is the result from the "target return" in your head? And how exactly do you plan to close that gap?
3. Graham published his report card from four judgments—1948, 1953, 1959, 1964—including the one that "subsequent experience proved was not a particularly good recommendation." Do you have a written record of your own judgments? If not, on what basis do you know the quality of your judgment?
4. He opposed starting a new dollar-cost averaging plan at a high level like 1964, and the reason was behavioral, not valuation-based: a plan that opens with a huge loss gets abandoned. Is your current investment plan designed around "how big a drawdown you can withstand without quitting," or around "the highest historical return"?
5. Zweig asks, "Why should the future return of stocks always be the same as their past return?" Point that question at the asset you have the most conviction in (some index, some industry, some country's market): of your reasons for believing in it, how much comes from mechanism, and how much merely from its price performance over the past few years?

Next episode: Chapter 4, "General Portfolio Policy: The Defensive Investor"—the rulers are in hand; now it's time to act. Graham lays out the famous 25%–75% band with the 50/50 baseline, and why periodic rebalancing turns the decision most easily hijacked by emotion into a cool piece of arithmetic. Zweig also translates it into a version you can execute today: the psychological hurdle of rebalancing is far harder than the mathematical one.